Home » Internet » Internet Marketing

Multi-variate Testing Made Simple

Nov 28, 2007
To run simple, easy and effective multi-variate tests on your web pages, you can use a tool such as Google's new and free Website Optimizer service. This tool lets you test various elements on your web page to see which ones have the most impact on your conversion rates.

For instance, by choosing to test three elements with two variations each, the number of "recipes" jumps to nine. With a multi-variate test, you can test up to 40 combinations. The greater the number of combinations you have, the longer it's going to take to achieve significant results because you have the traffic seeing lots of different pages, so you need lots of visitors to get numbers that mean something. But by using a "full factorial design" within the test, the experiment stops when it reaches a statistically valid conclusion.

It records the number of samples and the "spread" of the different elements. The spread is the relative difference in conversion rates between one element or recipe and another. The bigger the spread, the higher the probability that the apparent winner is the actual winner.

Once the experiment of recipes stops, the test is over. You can't turn it back on to continue to collect data running those particular combinations of elements to increase your test sample. The program will test all the recipes as equally as possible throughout the test period.

A program like Website Optimizer makes multi-variate testing extremely simple. Even an amateur with no technical savvy can have a test set up within a matter of hours. Also, having everything on one page rather than sending traffic to different pages actually simplifies the testing and tracking process.

The great thing about multi-variate testing is that you can test the way different elements interact with each other, something that is impossible to do with an A/B test. With this method, you can combine a variety of small improvements to create a huge improvement over time.

A weakness of the Website Optimizer program is that it's not good at testing pages that are radically different. To do that, try an A/B test to determine which page out pulls the other, then tweak the winner of the test.

Each recipe you develop provides specific information about each test variable. By analyzing the recipes together, you can determine the real effect of each variable as well as the interactions between the variables.

Using multi-variate matrix tests lets you increase the power of your tests, learn more in less time and thus increase your conversion rates at a much higher pace so you continually grow your customer base. Don't be afraid to test bold changes, and do be certain to focus your tests on the most important elements that relate to your conversion percentages. And every once in a while, question yourself and your assumptions to really challenge your tests.
About the Author
Glen Hopkins specializes in teaching struggling entrepreneurs how to turn their small Online businesses into thriving money machines all while working less and earning more. To get more information, including Free Reports, Videos and CDs, visit: http://www.GlenHopkins.name
Please Rate:
(Average: Not rated)
Print Email Report Share
Article Categories